Research map / Label-efficient and robust learning / Recognition and long-tail
Multi-dimensional Neural Decoding with Orthogonal Representations for Brain-Computer Interfaces
Kaixi Tian, Shengjia Zhao, Yuhan Zhang, Shan Yu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add OrthoSchema (orthogonal loss + session head + selective transfer) to RNN backbone | RNN · CO cross-session 2-shot | Direction accuracy (%): 72.60 -> 80.23 | Improved |
| add OrthoSchema to CNN backbone | CNN · CO cross-session 5-shot | Hand acceleration R2: 0.562 -> 0.550 | Hurt |
| add OrthoSchema to RNN backbone | RNN · CO cross-subject 2-shot | Direction accuracy (%): 45.53 -> 60.01 | Improved |
| add OrthoSchema to RNN backbone | RNN · CO to RT cross-paradigm 2-shot | Position R2: 0.579 -> 0.622 | Improved |
| add OrthoSchema to MLP backbone | MLP · CO cross-session 5-shot | Direction accuracy (%): 94.02 -> 94.39 | No clear effect |
| add OrthoSchema to MLP backbone | MLP · CO to RT cross-paradigm 5-shot | Position R2: 0.618 -> 0.624 | No clear effect |
| add orthogonal loss to base multi-task RNN (ablation) | RNN · CO cross-session 2-shot | Direction accuracy: about +6% | Improved |
| add session classification head to base multi-task RNN (ablation) | RNN · CO cross-session 2-shot | Direction accuracy: about +2% | Improved |
| combine orthogonal loss and session head (full method) vs base (ablation) | RNN · CO cross-session 2-shot | Direction accuracy: about +8% | Improved |
| vary orthogonal loss weight (0.001-0.005 or 0.1-0.5 vs 0.05) (ablation) | OrthoSchema RNN · CO cross-session 2-shot | Direction accuracy | Mixed |
| vary transfer feature retention ratio (10-20% or 50-100% vs 30%) (ablation) | OrthoSchema RNN · CO cross-session 2-shot | Direction accuracy | Mixed |
| add OrthoSchema vs multi-task baselines (UW, GradNorm) and neural decoders (LFADS, NoMAD) | RNN · CO cross-session 2-shot | Direction accuracy (%): 67.74 (NoMAD) -> 80.23 | Improved |
Closest papers by meaning
- SPINT: Spatial Permutation-Invariant Neural Transformer for Consistent Intracortical Motor Decoding (NeurIPS 2025 · similarity 0.88)
- Few-shot Algorithms for Consistent Neural Decoding (FALCON) Benchmark (NeurIPS 2024 · similarity 0.87)
- A Scalable, Causal, and Energy Efficient Framework for Neural Decoding with Spiking Neural Networks (NeurIPS 2025 · similarity 0.85)
- Generalizable, real-time neural decoding with hybrid state-space models (NeurIPS 2025 · similarity 0.85)
- CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding (NeurIPS 2025 · similarity 0.85)
- Neural decoding from stereotactic EEG: accounting for electrode variability across subjects (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
